An article that claims that /j/ and /i/ are phonemically the same and distinguished from each other by being syllabic or not is either very confused about the concept of "phoneme", or doesn't know the difference between "phonemically" and "segmentally".
The standard late-SPE analysis of the distinction is that [j,w] are [-syllabic] and [i,u] are [+syllabic]: they have a difference in a feature. With the advent of CV phonology in Clements & Keyser 1983, the feature syllabic was exiled in favor of some kind of suprasegmental distinction, such as "is dominated by C vs V", or "Is dominated by 'N' (nucleus)", and then things got more complex in moraic theory. In fact, in moraic theory, "consonantal" was redefined to essentially mean "non-syllabic" (j, w, h are [+consonantal]), which is okay because as argues by Hume & Odden, there is no good evidence for the feature consonantal (as defined in SPE).
Under the CV and later but not moraic accounts, the difference is not in terms of a segmental feature, it is in terms of a prosodic property. If one incorrectly things of "phonemes" as being defined solely in terms of segmental features, it would be true that [j,i] and [w,u] are featurally identical. (Indeed, moraic theorists do not uniformly embrace the "let's call them [+consonantal])" account of glides a.k.a semivowels.
The question of being "phonemic" is a question about phonological contrast, and not about phonetic properties. There is great phonetic similarity between [i,u] and [j,w], but that doesn't mean that [j] and [i] phonemically contrast. It is an open question whether there is a real, hard-core contrast between glides and vowels (and much ink spilled on the topic).
Although I don't accept this analysis as correct, the examples "year" versus "ear" can be phonemicized as /iir/ vs. /ir/, so there is no phonemic contrast. The rule is (as a first approximation) that high vowels become glides before vowels. Those who deny that there is a phonemic contrast between vowels and glides must analyze all glides as coming from corresponding vowels, by some rule. (Logically, they could also hold that all high vowels come from glides, but nobody does that).
Levin in her dissertation "A metrical theory of syllabicity", and subsequent post-CV phonology theories have sort of eliminated the question of vowel / glide contrast from serious consideration, because the distinction is purely predictable from prosodic position: being the head of the nucleus means you are a vowel, otherwise you are a glide. The discussion then has to move to the issue of nuclear vs. non-nuclear status.
As for the phonetics of glides vs. high vowels, there's sort of a mixed bag out there. The two main tendencies are that a glide is shorter than the corresponding high vowel, and glides are more constricted. Sometimes you get other differences, such as that in Logoori, /j/ is (typically produced as, though optionally) a very fronted tongue-tip down approximant, unlike /i/.
